INF-HatEval at SemEval-2019 Task 5: Convolutional Neural Networks for Hate Speech Detection Against Women and Immigrants on Twitter

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 420–425 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 420 INF-HatEval at SemEval-2019 Task 5: Convolutional Neural Networks for Hate Speech Detection Against Women and Immigrants on Twitter Alison P. Ribeiro Institute of Informatics Federal University of Goi´as Goiˆania – Goi´as – Brazil alisonrib17@gmail.com N´adia F. F. da Silva Institute of Informatics Federal University of Goi´as Goiˆania – Goi´as – Brazil nadia@inf.ufg.br Abstract In this paper, we describe our approach to detect hate speech against women and immi- grants on Twitter in a multilingual context, En- glish and Spanish. This challenge was propo- sed by the SemEval-2019 Task 5, where par- ticipants should develop models for hate spe- ech detection, a two-class classification where systems have to predict whether a tweet in En- glish or in Spanish with a given target (women or immigrants) is hateful or not hateful (Task A), and whether the hate speech is directed at a specific person or a group of individuals (Task B). For this, we implemented a Convolutio- nal Neural Networks (CNN) using pre-trained word embeddings (GloVe and FastText) with 300 dimensions. Our proposed model obtai- ned in Task A 0.488 and 0.696 F1-score for English and Spanish, respectively. For Task B, the CNN obtained 0.297 and 0.430 EMR for English and Spanish, respectively. 1 Introduction With the growth of users in social networks, there was also an increase in the odious activi- ties that permeate these communicative structures. According to Nockleby et al. (2000), hate speech can be defined as any communication that depre- cates a person or a group based on some charac- teristics such as race, color, ethnicity, gender, na- tionality, religion or other features. And the main motive that encourages users to spread hate on so- cial networks is anonymity, so users can spread hate words to a particular target. For this reason, the hatred propagated can generate irreversible consequences, where young people who approach with cyberbullying and homophobia, mainly, com- mit suicide. Nowadays, social networks like Twitter1, Fa- 1https://twitter.com/ cebook2 and YouTube3 are pressured to develop tools to fight the proliferation of hate in their networks. A good example of this is the German government that threatened to fine social networks by up to 50 million euros if they did not fight the spread of hate (Gamb¨ack and Sikdar, 2017). However, while there is plenty of available con- tent on social networks, the task of detecting hate speech remains difficult, largely because of the use of different sets of data for work, lack of bench- marking, and efficient approaches. Waseem, for example, bring a study focused on the detection of racism and sexism, whereas Nahar et al. 2012 and Sanchez and Kumar 2011) conducted a sur- vey on detecting bullying. For the detection of ho- mophobia, misogyny and xenophobia, the number of papers is still limited, one can cite a recent pa- per (Sanguinetti et al., 2018), where the authors sought to identify hate …

Content truncated to 500 words